Fix min/max value in Iceberg writes

Author: ianton-ru

src/Storages/ObjectStorage/DataLakes/Iceberg/IcebergWrites.cpp

@@ -1232,11 +1232,6 @@ void IcebergStorageSink::consume(Chunk & chunk)
         {
             auto [data_filename, data_filename_in_storage] = filename_generator.generateDataFileName();
             data_filenames[partition_key] = data_filename;
-            if (!statistics.contains(partition_key))
-            {
-                statistics.emplace(partition_key, current_schema->getArray(Iceberg::f_fields));
-            }
-            statistics.at(partition_key).update(part_chunk);
 
             auto buffer = object_storage->writeObject(
                 StoredObject(data_filename_in_storage), WriteMode::Rewrite, std::nullopt, DBMS_DEFAULT_BUFFER_SIZE, context->getWriteSettings());
@@ -1252,6 +1247,12 @@ void IcebergStorageSink::consume(Chunk & chunk)
                 configuration->getFormat(), *write_buffers[partition_key], *sample_block, context, format_settings);
         }
 
+        if (!statistics.contains(partition_key))
+        {
+            statistics.emplace(partition_key, current_schema->getArray(Iceberg::f_fields));
+        }
+        statistics.at(partition_key).update(part_chunk);
+
         writers[partition_key]->write(getHeader().cloneWithColumns(part_chunk.getColumns()));
     }
 }

tests/integration/test_storage_iceberg/test.py

@@ -3744,3 +3744,18 @@ def test_system_tables_partition_sorting_keys(started_cluster, storage_type):
     """).strip().lower()
 
     assert res == '"bucket(16, id), day(ts)","id desc, hour(ts) asc"'
+
+
+def test_iceberg_write_minmax(started_cluster):
+    instance = started_cluster.instances["node1"]
+    TABLE_NAME = "test_iceberg_write_minmax_" + get_uuid_str()
+
+    create_iceberg_table("local", instance, TABLE_NAME, started_cluster, "(x Int32, y Int32)", partition_by="identity(x)")
+
+    instance.query(f"INSERT INTO {TABLE_NAME} VALUES (1, 1), (1, 2)", settings={"allow_experimental_insert_into_iceberg": 1})
+
+    res = instance.query(f"SELECT x,y FROM {TABLE_NAME} WHERE y=1 ORDER BY ALL").strip()
+    assert res == "1\t1"
+
+    res = instance.query(f"SELECT x,y FROM {TABLE_NAME} WHERE y=2 ORDER BY ALL").strip()
+    assert res == "1\t2"